Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method for managing a plurality of queues of a network interface card (NIC) of a host computer, the method comprising: defining first, second, and third subsets of the queues, wherein the first subset of queues is associated with a first feature for aggregating multiple related data messages received by the NIC into a single data message, the second subset of queues is associated with a second feature for load balancing data messages received by the NIC across multiple queues of the associated subset of queues, and the third subset of queues is associated with both the first and second features; receiving a request from a data compute node (DCN) executing on the host computer for data messages addressed to the DCN to be processed with both the first feature for aggregating multiple related data messages into a single data message and second feature for load balancing data messages across multiple queues; and configuring the NIC to direct data messages received for the DCN to a particular queue selected from the third subset of queues.
This invention relates to managing multiple queues in a network interface card (NIC) of a host computer to optimize data processing for a data compute node (DCN). The problem addressed is efficiently handling data messages by combining aggregation and load balancing features in a flexible queue management system. The method involves defining three subsets of queues: a first subset for aggregating multiple related data messages into a single message, a second subset for distributing messages across multiple queues to balance the load, and a third subset that supports both features. When a DCN requests processing with both aggregation and load balancing, the NIC is configured to route incoming messages for that DCN to a queue from the third subset. This ensures messages are both aggregated and distributed efficiently, improving performance and resource utilization. The solution dynamically assigns queues based on the required features, allowing adaptable handling of different data processing needs.
2. The method of claim 1 further comprising, upon receiving the request from the DCN, initially configuring the NIC to direct data messages received for the DCN to a default queue that is not in any of the first, second, and third subsets of queues.
A method for managing network traffic in a data center network (DCN) involves dynamically allocating network interface card (NIC) queues to optimize data flow. The method addresses inefficiencies in traditional queue management systems, where static configurations fail to adapt to varying traffic demands, leading to congestion or underutilization. The invention dynamically assigns NIC queues to different subsets based on traffic characteristics, improving throughput and reducing latency. Initially, the NIC is configured to direct incoming data messages for the DCN to a default queue outside the predefined subsets. This default queue acts as a temporary holding area until the system evaluates traffic patterns and reallocates queues as needed. The method ensures that traffic is efficiently distributed across the NIC's available resources, preventing bottlenecks and enhancing overall network performance. By dynamically adjusting queue assignments, the system adapts to real-time demands, optimizing resource utilization and maintaining high-speed data transmission. The solution is particularly useful in high-traffic environments where static configurations would otherwise degrade performance.
3. The method of claim 2 , wherein configuring the MC to direct data messages received for the DCN to the particular queue comprises: monitoring a load on the default queue; and when the load on the default queue exceeds a first threshold, configuring the NIC to direct subsequent data messages received for the DCN to the particular queue.
This invention relates to network traffic management in data center networks (DCNs), specifically addressing the problem of efficiently distributing incoming data messages to prevent congestion in default queues. In a DCN, network interface cards (NICs) receive data messages destined for various endpoints, which are typically routed to a default queue for processing. However, when the load on this default queue exceeds a predefined threshold, performance degradation or packet loss can occur. The invention provides a method to dynamically redirect incoming data messages to a particular queue when the default queue becomes overloaded. The method involves monitoring the load on the default queue in real-time. When the monitored load exceeds a first threshold, the NIC is reconfigured to direct subsequent data messages intended for the DCN to the particular queue instead of the default queue. This redistribution helps balance the load and prevents congestion, ensuring smoother network operation. The particular queue may be a secondary or dedicated queue with sufficient capacity to handle the redirected traffic. The method may also include reverting to the default queue once the load decreases below a second threshold, maintaining optimal traffic distribution. This approach enhances network efficiency and reliability in high-traffic scenarios.
4. The method of claim 2 , wherein the particular queue is a first queue, wherein configuring the NIC to direct data messages received for the DCN to the first queue comprises: monitoring loads on the default queue and the queues of the first, second, and third subsets; when the load on the default queue exceeds a first threshold, configuring the NIC to direct subsequent data messages received for the DCN to a second queue in the first subset of queues; and when the load on the second queue exceeds a second threshold, configuring the NIC to direct subsequent data messages received for the DCN to the first queue.
In data center networking, managing network traffic efficiently is critical to prevent congestion and ensure low-latency communication. A method dynamically adjusts data message routing in a network interface card (NIC) to balance workloads across multiple queues. The NIC initially directs incoming data messages for a distributed control network (DCN) to a default queue. The system monitors the load on the default queue and multiple subsets of queues. If the load on the default queue exceeds a first threshold, the NIC is reconfigured to redirect subsequent data messages to a second queue in a first subset of queues. If the load on the second queue then exceeds a second threshold, the NIC is further reconfigured to redirect subsequent data messages back to the first queue. This dynamic redistribution ensures that traffic is evenly distributed, preventing bottlenecks and improving overall network performance. The method leverages real-time load monitoring to optimize queue utilization, enhancing efficiency in high-traffic scenarios.
5. The method of claim 1 , wherein configuring the NIC to direct the data messages to the particular queue comprises defining a filter for the DCN and configuring the NIC to apply the filter to the particular queue.
A method for optimizing data message processing in a networked computing environment involves configuring a network interface card (NIC) to direct data messages to a specific queue within a data center network (DCN). The method addresses inefficiencies in data routing by dynamically assigning messages to queues based on predefined criteria, reducing latency and improving throughput. The NIC is programmed with a filter that identifies and categorizes incoming data messages according to their characteristics, such as source, destination, or protocol type. Once the filter is defined, the NIC applies it to route the messages to the appropriate queue, ensuring that messages are processed in an orderly and efficient manner. This approach minimizes the overhead of traditional software-based routing by offloading the filtering and queuing tasks to the NIC hardware, which operates at higher speeds. The method is particularly useful in high-performance computing environments where low-latency and high-throughput data processing are critical. By leveraging hardware acceleration, the system enhances overall network performance while reducing the computational burden on the host processor. The filter can be dynamically updated to adapt to changing network conditions or traffic patterns, ensuring sustained efficiency. This technique is applicable in cloud computing, data centers, and other environments where large-scale data processing is required.
6. The method of claim 5 , wherein data messages received by the NIC match the filter based on at least one of an outer destination address, an inner destination address, and a network identifier.
A method for filtering data messages in a network interface controller (NIC) involves processing incoming data messages to determine whether they match predefined filter criteria. The filter criteria are based on at least one of an outer destination address, an inner destination address, or a network identifier. The outer destination address refers to the destination address in the outer header of a tunneled packet, while the inner destination address refers to the destination address in the inner header of the tunneled packet. The network identifier may include a virtual local area network (VLAN) tag or other network segmentation identifier. The NIC evaluates each incoming data message against these criteria to determine whether the message should be processed further or discarded. This filtering mechanism allows the NIC to efficiently manage network traffic by selectively processing only relevant messages, reducing unnecessary processing overhead and improving overall network performance. The method is particularly useful in environments where multiple network segments or virtual networks coexist, requiring precise traffic filtering to ensure proper routing and security.
7. The method of claim 1 , wherein the NIC implements the first feature for aggregating multiple related data messages received by the NIC into a single data message and the second feature for load balancing data messages received by the NIC across multiple queues of the associated subset of queues.
A network interface controller (NIC) is configured to enhance data processing efficiency by implementing two key features. The first feature involves aggregating multiple related data messages received by the NIC into a single data message, reducing the overhead associated with handling individual messages. This aggregation process ensures that related data is grouped together, minimizing the number of discrete operations required for processing. The second feature involves load balancing the received data messages across multiple queues within a subset of available queues. This distribution ensures that the workload is evenly spread, preventing bottlenecks and improving overall system performance. The NIC dynamically assigns incoming data messages to different queues based on factors such as message priority, source, or content, optimizing resource utilization. By combining these features, the NIC efficiently manages high volumes of data while maintaining low latency and high throughput. This approach is particularly beneficial in high-performance computing environments where rapid and balanced data processing is critical. The system ensures that data is processed in an organized and efficient manner, enhancing the overall performance of the network infrastructure.
8. The method of claim 1 , wherein virtualization software of the host computer implements the first feature for aggregating multiple related data messages received by the NIC into a single data message and the NIC implements the second feature for load balancing data messages received by the NIC across multiple queues of the associated subset of queues.
In the field of network data processing, this invention addresses inefficiencies in handling high volumes of data messages in virtualized computing environments. The technology focuses on optimizing data flow between a network interface card (NIC) and a host computer running virtualization software. The system combines two key features to improve performance. First, the virtualization software aggregates multiple related data messages received by the NIC into a single data message, reducing overhead and processing load. Second, the NIC itself distributes incoming data messages across multiple queues within a designated subset of queues, ensuring balanced workload distribution and preventing bottlenecks. This dual approach enhances throughput and reduces latency by leveraging both hardware and software optimizations. The aggregation feature minimizes the number of operations required to process related messages, while the load-balancing feature ensures even distribution of tasks across available resources. Together, these mechanisms improve efficiency in virtualized environments where multiple virtual machines or containers share network resources. The solution is particularly valuable in high-performance computing scenarios where minimizing latency and maximizing throughput are critical.
9. The method of claim 1 , wherein: the second subset of queues are logical queues; each logical queue in the second subset of queues is associated with a plurality of hardware queues; and the second feature balances data message traffic for a single DCN across multiple hardware queues.
This invention relates to data center networking (DCN) systems, specifically addressing the challenge of efficiently distributing data message traffic across multiple hardware queues to optimize performance and resource utilization. The method involves managing a plurality of queues, including a first subset of queues and a second subset of queues, where the second subset consists of logical queues. Each logical queue in the second subset is linked to multiple hardware queues, enabling traffic distribution across these hardware resources. The method further includes a feature that balances data message traffic for a single DCN across the multiple hardware queues, ensuring efficient load distribution and preventing bottlenecks. This approach enhances network performance by dynamically allocating traffic to available hardware queues, improving throughput and reducing latency in data center environments. The system dynamically assigns and manages these queues to adapt to varying traffic conditions, ensuring optimal utilization of network resources. The method is particularly useful in high-traffic scenarios where efficient traffic distribution is critical for maintaining network efficiency and reliability.
10. The method of claim 1 , wherein if there are no queues for a requested feature available, a new subset of queues is defined for the requested feature.
This invention relates to a system for managing and allocating queues in a computing environment, particularly for handling requests for various features or services. The problem addressed is the efficient distribution of requests to available queues to ensure optimal resource utilization and minimize delays. The system dynamically assigns requests to existing queues based on availability and performance metrics. If no suitable queues are available for a requested feature, the system automatically creates a new subset of queues specifically for that feature. This ensures that new requests are processed without delay, even if existing queues are fully allocated. The method involves monitoring queue status, evaluating request types, and dynamically adjusting queue allocations to maintain system efficiency. The invention improves scalability and responsiveness by preventing bottlenecks caused by insufficient queue resources. The dynamic creation of new queues ensures that system performance remains consistent even under varying workload conditions. This approach is particularly useful in high-demand environments where features may experience sudden spikes in request volume. The system avoids manual intervention by autonomously managing queue resources, reducing administrative overhead and improving overall system reliability.
11. The method of claim 10 , wherein queues in the new subset are reallocated from at least one existing subset of queues.
This invention relates to dynamic queue management in computing systems, particularly for optimizing resource allocation in distributed or parallel processing environments. The problem addressed is inefficient resource utilization when static queue configurations fail to adapt to changing workload demands, leading to bottlenecks or underutilization. The method involves dynamically reallocating queues from one subset of queues to another to improve performance. A subset of queues is identified for reallocation based on criteria such as workload imbalance, resource contention, or performance metrics. Queues from an existing subset are then redistributed to a new subset to balance the load or optimize resource usage. This reallocation can involve migrating tasks, adjusting priorities, or reconfiguring queue assignments to better match current system demands. The method ensures minimal disruption by maintaining queue integrity during the transition and may involve preemptive or adaptive reallocation strategies. The goal is to enhance system efficiency by dynamically adapting queue structures to varying workloads, reducing latency, and improving throughput.
12. A non-transitory machine readable medium storing a program which when executed on set of processing units of a host computer manages a plurality of queues of a network interface card (NIC) of the host computer, the program comprising sets of instructions for: defining first, second, and third subsets of the queues, wherein the first subset of queues is associated with a first feature for aggregating multiple related data messages received by the NIC into a single data message, the second subset of queues is associated with a second feature for load balancing data messages received by the NIC across multiple queues of the associated subset of queues, and the third subset of queues is associated with both the first and second features; receiving a request from a data compute node (DCN) executing on the host computer for data messages addressed to the DCN to be processed with both the first feature for aggregating multiple related data messages into a single data message and second feature for load balancing data messages across multiple queues; and configuring the NIC to direct data messages received for the DCN to a particular queue selected from the third subset of queues.
This invention relates to network interface card (NIC) management in host computers, specifically addressing the need for efficient handling of data messages in high-performance computing environments. The system manages multiple queues on a NIC to optimize data processing for different workloads. The queues are divided into three subsets: the first subset supports message aggregation, combining multiple related data messages into a single message to reduce overhead; the second subset enables load balancing, distributing incoming data messages across multiple queues to prevent bottlenecks; and the third subset combines both features, allowing aggregated messages to be load-balanced. The system processes requests from data compute nodes (DCNs) running on the host computer, configuring the NIC to route messages for a DCN to a queue in the third subset when both aggregation and load balancing are required. This ensures efficient data handling by leveraging the appropriate NIC features based on the workload demands. The solution improves performance by dynamically assigning queues to optimize throughput and reduce latency in network communications.
13. The non-transitory machine readable medium of claim 12 , wherein the program further comprises a set of instructions for initially configuring the NIC, upon receiving the request from the DCN, to direct data messages received for the DCN to a default queue that is not in any of the first, second, and third subsets of queues.
A system and method for managing network traffic in a data center involves a network interface card (NIC) configured to handle data messages for a data center network (DCN). The NIC includes multiple queues for processing data messages, divided into subsets for different priority levels or functions. When the DCN sends a request to the NIC, the NIC initially directs incoming data messages for the DCN to a default queue that is separate from the predefined subsets of queues. This default queue ensures that data messages are temporarily stored or processed in a neutral queue before being assigned to one of the specialized subsets. The system may also include instructions for dynamically reconfiguring the NIC to adjust queue assignments based on network conditions or DCN requirements. The approach improves traffic management by preventing congestion in priority queues and ensuring efficient data handling. The solution is particularly useful in high-performance computing environments where network latency and throughput are critical.
14. The non-transitory machine readable medium of claim 13 , wherein the set of instructions for configuring the NIC to direct data messages received for the DCN to the particular queue comprises a set of instructions for: monitoring a load on the default queue; and when the load on the default queue exceeds a first threshold, configuring the NIC to direct subsequent data messages received for the DCN to the particular queue.
A system and method for managing network traffic in a data center network (DCN) involves dynamically distributing incoming data messages to optimize performance. The technology addresses the problem of congestion in default network queues, which can lead to bottlenecks and degraded network efficiency. The solution includes a network interface card (NIC) configured to monitor the load on a default queue. When the load exceeds a predefined threshold, the NIC automatically redirects subsequent data messages intended for the DCN to a designated alternative queue. This dynamic redirection helps balance the workload across multiple queues, preventing overload on the default queue and improving overall network throughput and latency. The system may also include additional logic to revert traffic distribution once the load on the default queue falls below a second threshold, ensuring efficient use of network resources. The approach leverages hardware-based monitoring and redirection to minimize software overhead and provide real-time traffic management.
15. The non-transitory machine readable medium of claim 13 , wherein the particular queue is a first queue, wherein the set of instructions for configuring the NIC to direct data messages received for the DCN to the first queue comprises sets of instructions for: monitoring loads on the default queue and the queues of the first, second, and third subsets; when the load on the default queue exceeds a first threshold, configuring the NIC to direct subsequent data messages received for the DCN to a second queue in the first subset of queues; and when the load on the second queue exceeds a second threshold, configuring the NIC to direct subsequent data messages received for the DCN to the first queue.
A system for managing network traffic in a data center network (DCN) involves dynamically distributing incoming data messages across multiple queues to balance load and prevent congestion. The system includes a network interface controller (NIC) configured to monitor traffic loads on a default queue and multiple subsets of queues. When the load on the default queue exceeds a first threshold, the NIC redirects subsequent data messages to a second queue in a first subset of queues. If the load on the second queue then exceeds a second threshold, the NIC further redirects traffic to a first queue in the same subset. This dynamic redistribution ensures efficient load balancing by leveraging multiple queues, preventing bottlenecks, and maintaining optimal network performance. The system may also include additional subsets of queues for further traffic management, with similar load-based redirection logic applied to those subsets. The NIC continuously monitors queue loads and adjusts message routing accordingly, ensuring adaptive traffic distribution in response to real-time network conditions. This approach enhances scalability and reliability in high-traffic environments by dynamically allocating resources based on demand.
16. The non-transitory machine readable medium of claim 12 , wherein the set of instructions for configuring the NIC to direct the data messages to the particular queue comprises a set of instructions for defining a filter for the DCN and configuring the NIC to apply the filter to the particular queue, wherein data messages received by the NIC match the filter based on at least one of an outer destination address, an inner destination address, and a network identifier.
A system for optimizing data message processing in a data center network (DCN) involves a network interface card (NIC) configured to direct incoming data messages to specific queues based on predefined filters. The NIC applies these filters to classify and route messages efficiently, reducing processing overhead. The filters are defined by criteria such as outer destination addresses, inner destination addresses, or network identifiers, ensuring precise message routing. This approach enhances network performance by minimizing unnecessary processing and improving data flow management within the DCN. The NIC's ability to dynamically apply these filters allows for adaptive traffic handling, supporting scalable and efficient network operations. The system is particularly useful in high-traffic environments where rapid and accurate message classification is critical for maintaining low latency and high throughput. By leveraging hardware-based filtering, the solution offloads processing tasks from the host system, further optimizing resource utilization. This method ensures that data messages are directed to the appropriate queues based on their attributes, streamlining network operations and improving overall system efficiency.
17. The non-transitory machine readable medium of claim 12 , wherein: the second subset of queues are logical queues; each logical queue in the second subset of queues is associated with a plurality of hardware queues; and the second feature balances data message traffic for a single DCN across multiple hardware queues.
The invention relates to data center networking (DCN) systems and addresses the challenge of efficiently distributing data message traffic across multiple hardware queues to improve performance and resource utilization. In a data center network, data messages are often processed through multiple queues to manage traffic flow and prioritize tasks. However, traditional systems may struggle to balance traffic effectively, leading to bottlenecks or underutilized resources. The invention provides a method for managing data message traffic in a DCN by using logical queues that map to multiple hardware queues. A subset of these queues are logical, meaning they are virtual constructs that can be dynamically assigned to different hardware queues. Each logical queue in this subset is linked to a plurality of hardware queues, allowing traffic to be distributed across them. This approach enables the system to balance data message traffic for a single DCN across multiple hardware queues, improving efficiency and reducing congestion. The system dynamically adjusts traffic distribution based on real-time conditions, ensuring optimal performance. This method enhances scalability and flexibility in data center networking by decoupling logical traffic management from physical hardware constraints.
18. The non-transitory machine readable medium of claim 12 , wherein the program further comprises a set of instructions for defining a new subset of queues for a requested feature if there are no queues available for the requested feature, wherein queues in the new subset are reallocated from at least one existing subset of queues.
This invention relates to a system for managing queues in a computing environment, specifically addressing the challenge of efficiently allocating and reallocating queues to support different features or tasks. The system dynamically adjusts queue assignments to optimize resource utilization when existing queues are unavailable for a requested feature. When no queues are available for a new feature request, the system creates a new subset of queues by reallocating queues from at least one existing subset. This reallocation ensures that resources are dynamically redistributed to meet current demands without requiring additional hardware or static configurations. The system includes a machine-readable medium storing instructions for performing these operations, enabling flexible and adaptive queue management. The solution improves system efficiency by avoiding idle resources and ensuring that queues are allocated based on real-time requirements, reducing bottlenecks and enhancing performance in computing environments where multiple features compete for limited queue resources. The reallocation process is automated, minimizing manual intervention and ensuring seamless operation. This approach is particularly useful in high-demand environments where features may have varying and unpredictable resource needs.
19. The non-transitory machine readable medium of claim 12 , wherein the NIC implements the first feature for aggregating multiple related data messages received by the NIC into a single data message and the second feature for load balancing data messages received by the NIC across multiple queues of the associated subset of queues.
A network interface controller (NIC) is configured to optimize data processing by implementing two key features. The first feature aggregates multiple related data messages received by the NIC into a single data message, reducing the overhead of handling individual messages. This aggregation is particularly useful in high-throughput environments where minimizing processing latency is critical. The second feature enables load balancing of received data messages across multiple queues associated with the NIC. By distributing messages across these queues, the NIC ensures efficient utilization of system resources and prevents bottlenecks that could arise from uneven workload distribution. The NIC dynamically assigns messages to the appropriate queues based on predefined criteria, such as message type, source, or priority, to maintain balanced performance. This dual functionality enhances overall system efficiency by reducing processing overhead and ensuring equitable resource allocation. The NIC operates independently of the host system's central processing unit (CPU), offloading these tasks to improve throughput and reduce latency. This approach is beneficial in data-intensive applications, such as cloud computing, high-frequency trading, or real-time analytics, where rapid and efficient data handling is essential. The NIC's capabilities are implemented in firmware or hardware, ensuring low-latency execution without relying on software-based solutions.
Unknown
June 16, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.